A generalized neural network model for the VLE of supercritical carbon dioxide fluid extraction of fatty oils

Abstract

In the present work, a neural network (NN) model was developed for the precise prediction of the phase behavior of supercritical carbon dioxide (SC-CO2) and fatty oils. A total of 678 SC-CO2 + fatty oils vapor–liquid equilibrium datasets for fatty acids, methyl and ethyl esters plus 120 data points for further accuracy examinations were used. The Cascade-Forward scheme was used as the basic architecture for the NN model calculations and predictions. Comparison between the values of the average absolute deviation of the NN model and the most important existing models showed that the NN model outperforms the other alternatives. The overall AAD for CO2 mole fractions in liquid and vapor phases were obtained to be 1.516 and 0.312%, respectively.